1

Strong reduction of V4+ amount in vanadium oxide/hexadecylamine

nanotubes by doping with Co2+ and Ni2+ ions: EPR and magnetic

studies M.E. Saleta a, H.E. Troiani a, S. Ribeiro Guevara a, R.D. Sánchez a*, M. Malta b and R.M.

Torresi c

a) Centro Atómico Bariloche, CNEA, (8400) S. C. de Bariloche (RN), Argentina.

b) Depto. de Cs. Exatas e da Terra, Univ. do Estado da Bahia, Cabula Salvador (BA), Brazil.

c) Instituto de Química, Universidade de São Paulo, São Paulo (SP), Brazil.

(29 June 2010)

Abstract

In this work we present a complete characterization and magnetic study of vanadium

oxide/hexadecylamine nanotubes (VOx/Hexa NT’s) doped with Co2+ and Ni2+ ions. The

morphology of the NT’s has been characterized by Transmission Electron Microscopy (TEM)

while the metallic elements have been quantified by Instrumental Neutron Activation Analysis

(INAA) technique. The static and dynamic magnetic properties were studied collecting data of

magnetization as a function of magnetic field and temperature and by Electron Paramagnetic

Resonance (EPR). We observed that the incorporation of metallic ions (Co2+, S=3/2 and Ni2+,

S=1) decrease notably the amount of V4+ ions in the system, from 14-16% (non-doped case) to

2-4%, with respect to the total vanadium atoms into the tubular nanostructure, improving

considerably their potential technological applications as Li-ion batteries cathodes.

Keywords: nanostructured materials; magnetization; electron paramagnetic resonance;

magnetic measurement; transmission electron microscopy

1. Introduction

Oxide nanotubes (NT’s) such as SiO2, TiO2, ZnO, VOx and so on, have been the subject of a

very active research due to their distinctive physical and chemical properties, which have

promising technological applications [1]. Among these compounds are the vanadium oxide

(VOx) NT’s, in which the walls are constituted by alternated layers of VOx and surfactant. The

surfactant or template helps to form the skeleton, which provides support and hardness to the

NT’s walls. Different templates have been used as primary monoamines (CnH2n+1NH2) and α,ω-

diamines (H2N(CH2)n)NH2) [2, 3]. The layers of VOx are constituted by V ions that, in this case,

* Corresponding author: Tel.: +54 2944 445158; fax: +54 2944 445299: e-mail address: rodo@cab.cnea.gov.ar (R.D. Sánchez)

2

can be in two different oxidation states (4+, with a magnetic spin S=1/2 and 5+, the diamagnetic

ions with S=0). The presence of V4+ ions in the estructure is essential for the rolling-up of the

VOx layer in order to form the NT [2]. More over, Vera-Robles and Campero [4] observed that

the presence of both V species (V4+ and V5+) is fundamental for the scrolling process that occurs

during hydrothermal treatment. They worked with vanadyl (IV) acetate and dodecylamine as

precursor and obtained the V mixed valence by oxidizing the V4+. In two previous works,

Krusin-Elbaum [5] and Vavilova [6] studied VOx/dodecylamine NT’s and they reported a high

percentage of V4+ ions detected by magnetic experiments. Even more, half of these magnetic

ions are antiferromagnetically coupled dimmers.

The oxide layer can be described using "V-O double layer" structure of the BaV7O16 · nH2O

(P42/m) [5]. The structure is shown in Fig. 1. In the V7O16 structure both V(1) and V(2) sites are

coordinated with six oxygen atoms (VO6-octahedral). These octahedra form the vanadium oxide

double layer. Between these layers, the remaining vanadium atoms are located in the V(3) sites

and they are coordinated with four oxygen atoms forming tetrahedrons [7].

The effect of cation exchange in VOx/surfactant NT’s was initially studied by Nesper’s

group [8] and subsequently by other authors [9,10]. They reported that some transition metals,

alkali metals and alkali-earths can be exchanged with good preservation of the tubular shape.

Azambre and Hudson have also synthesized copper nanoparticles (NP’s) within

VOx/dodecylamine NT’s [11] and our group has studied nanocomposites constituted by FeOy

NP’s and VOx/hexadecylamine NT’s [12].

The VOx NT’s present several potential technological applications, such as sensing

elements [13], catalysis [14] and Li-ion batteries cathodes [15]. In the last case, it was found

that the amount of V4+ in the structure is critical as it affects and reduce notabily the

performance of the cathode.

In this context, we present a complete structural, magnetic and Electron Paramagnetic

Resonance (EPR) study on VOx/hexadecylamine nanotubes in which magnetic Co or Ni ions are

intercalated in the structure. Our experiments confirm that the incorporation of Co2+ and Ni2+

ions into the amine layers reduces notably the amount of the isolated and paramagnetic V4+ ions.

2. Experimental

The VOx/Hexa NT’s were synthesized in two steps. In the first one, 3.09g of crystalline

V2O5 (Aldrich) were slowly added in a beaker containing an ethanolic solution (5.7mL) of

hexadecylamine 90% (4.55g, Aldrich). The mixture was then maintained under magnetic

stirring during two hours and, after that, 16.7mL of deionized water were added and the material

was aged during 48h. The second step consists in a hydrothermal digestion of the gel during 7

3

days at 180°C in order to form the non-doped NT’s. Afterwards, to obtain the doped NT’s, the

VOx/Hexa nanostructures were introduced into a solution if ethanol/water (4:1 v/v) with the

metallic ions, where a fraction of the amines were exchanged [8, 16]. We used CoCl2.6H2O

(Merck) and NiCl2.6H2O (Merck) in each solution, respectively, in a molar ratio excess of 4/1

(metallic cation/V2O5-hexadecylamine). The solution was stirred for 3 hours and filtered, and

the solid black product was rinsed and dried under vacuum.

The concentration of V and Co in the NT’s was determined by Instrumental Neutron

Activation Analysis (INAA). The samples were irradiated in the RA-6 research nuclear reactor,

located at Centro Atómico Bariloche (Bariloche Atomic Center - Argentina), in a thermalized

neutron flux (φth≅8×1011 n.cm-2.s-1). Vanadium and cobalt concentrations were determined by

comparison with high purity metallic standard materials (V: Johnson Matthey, 99.7% purity,

and Co: R/X Reactor Experiments, 99.95% purity). Vanadium was measured by evaluating the

1.43 MeV emission of the 52V isotope (half-life of 3.743 minutes), while cobalt concentrations

were determined by the 1.17 and 1.33 MeV emissions of 60Co (half-life of 1925.28 days).

Gamma-ray spectra were collected after appropriate decaying times regarding activation

products half lives. In the case of Co, the decay time was adjusted to allow complete decay of 60Com (half-life of 10.467 minutes) to the ground state for the samples and standards. Due to the

long half-life of the 58Ni(n,p) 58Co reaction (half-life of 70.86 days) we did not quantify the

amount of Ni with this technique.

The tubular shape of the products was confirmed by Transmission Electron Microscopy

(TEM) and the local composition was characterized by EDS. Both studies were performed in a

CM 200 Philips microscope (LaB6 cathode, 200 keV). The X-Ray Diffraction (XRD) patterns

were acquired in a Philips PW 1700 diffractometer using an aluminum sample holder and CuKα

radiation (1.54186 Å).

EPR spectra were collected with a Bruker ESP-300 spectrometer, operating at X-band (9.5

GHz) varying the temperature between 100K and 300K. Two additional spectra at K-band (24

GHz) and Q-band (35 GHz) were taken at room temperature to complete the energy levels

involved in the magnetic transitions.

The static magnetic characterization of the samples was performed in a commercial

superconducting quantum interference device magnetometer (Quantum Design MPMS-5S and

MPMS-XL) with fields up to 70kOe. The measurements as a function of temperature were

performed in the 5K-360K range.

3. Results

3.1 Morphology and composition

4

In order to understand the shape formation mechanism of the NT’s, we show in Fig. 2a and

2b TEM images of tubes that have not completely closed. Schematic representations of the

rolling-up of planes are presented in Fig. 2c and 2d. These photographs confirm the previous

rolling hypothesis mentioned by Nesper and Muhr [17].

A typical TEM image of non-doped NT is presented in Fig. 3a. The dark and bright lines are

associated to the alternated layers of VOx and Hexa showing the multiwall characteristics of

these NT’s, present in all the studied samples. Comparing the doped and the non-doped NT’s,

we observe an increment of defects in the doped samples, like loss of definition and waving

surface of the walls. These defects can be a consequence of the doping process when the Hexa

was partially exchanged by the metallic ions. In the middle picture (Fig. 3b) the EDS spectra of

the samples is shown. In the Co and Ni doped NT’s, an extra peak can be clearly detected due to

the presence of each doping element. The V and Co amounts, determined by INAA, are

presented in Table 1. The atomic mol ratio quantification between Co and V using both

techniques, local (EDS) and global (INAA), are in good agreement, yielding the value Co/V =

0.14. With the EDS technique we obtain 0.16 for the Ni/V ratio.

The XRD patterns are presented in Fig. 3c; we can considerer two regions in the patterns:

the low angle (LA, 1.5º ≤ 2θ ≤ 15º) and high angle (HA, 15º ≤ 2θ) region. The LA data have the

reflections indexed as 00l, which provide information about the interlamellar distance (d). On

the other hand the HA measurements have the hk0 reflections, which provide the VOx cell

parameters; these last peaks have lower intensity than the LA peaks. In order to index the cell

we assumed a planar square lattice [2]. The interlamellar distances decrease with doping, which

is in agreement with the exchange of Hexa by Ni or Co ions. The values of d are presented in

Table 1. The hk0 reflections do not vary with the incorporation of Ni and Co, which indicates

that the samples have practically the same cell parameters. This result is an indication that the

metallic ions are not incorporated to the VOx double layered structure. In this way, we assume

that either the doped ions are incorporated to the structure or they replace the amines in the

skeleton of the NT’s. The calculated cell parameter is a ≈ 5.93(2)Å for the three samples.

3.2. EPR measurements

The X-band EPR spectra measured at room temperature are shown in Fig 4. From the

bottom to the upper plot, the resonance lines for the three studied samples correspond to: non-

doped (only magnetic vanadium ions), Co-doped (magnetic vanadium and cobalt ions) and Ni-

doped (magnetic vanadium and nickel ions).

The three X-band EPR spectra can be described by the following Hamiltonian:

+++ =+= 22 ,NiCoiV HHH 4 , (1)

5

where +4H Vis the Hamiltonian given by the V4+ ions and ++= 22 ,NiCoiH is the contribution to

the spin Hamiltonian of the transition metal doped ion (i=Co2+ and Ni2+). In the non-doped

sample only the Hamiltonian of the V4+ contributes to the resonance, while for the doped

samples we have to include the extra term of metal dopants to describe the absorption.

The non-doped sample can be adjusted with a single line centered at g≈1.96 and that can be

associated with the V4+ (S=1/2) ions. However, when the spectra are measured at higher

frequencies (K and Q bands), the lines become asymmetric (different heights between the

maximum -or minimum- and the base line). This asymmetry means that the sample presents

crystalline anisotropy. The Hamiltonian which describes this anisotropy is presented in Eq. 2.

This expression also includes the hyperfine interaction between the electronic and nuclear spin

of 51V (abundance 99.76%, I=7/2) [18].

)]..()..[()]..()..[( |||| yyxxzzyyxxzzBV ISISAISASHSHgSHg +++++= ⊥⊥+ μ4H (2)

where μB is the Bohr magneton, Sj and Ij (j = x, y, z) are the projections in the j-direction of the

electronic spin and nuclear spin operator respectively; g|| and g⊥ are the parallel and

perpendicular (respect to magnetic field - H) components of the g-factor, and finally A|| and A⊥

are the principal (parallel and perpendicular) components of the hyperfine tensor in magnetic

field units – G.

In order o fit correctly the EPR spectra it was necessary to considerer a random distribution

of crystalline orientations (powder distribution) and we only took into account the Zeeman term

(first term of the Eq. 2). The experimental and theoretical fittings are presented in Fig. 5. The

obtained g components of the non-doped NT’s are shown in Table 2. Now, when we subtracte

the calculated powder line from the experimental resonance, a low intensity spectrum formed by

equally spaced lines remains. These absorptions correspond to the hyperfine structure (hfs)

produced by the interaction between the electronic and nuclear spins of 51V (both terms of Eq.

2). Contrary to those vanadium ions that produce the subtracted powder line, the hfs is an

indication that some V4+ ions are practically isolated without interacting with other V4+ ions. In

the case of the V ions which contribute to the powder line, the hfs collapses into a single line

(width 150G, X-band). This effect can be explained by the exchange interaction between near

V4+ ions (exchange-narrowing). This effect was studied by Deigen and co-workers in Mn2+ ions

[19] and by Wiench et al. in V4+ ions [20].

6

In order to describe the 51V4+ EPR in our experiments, the magnetic field positions of the hfs

satellites lines can be expressed as [18, 21]:

( ) ⎟⎠⎞

⎜⎝⎛+⎟

⎠⎞

⎜⎝⎛ −−−⎟⎟

⎠

⎞⎜⎜⎝

⎛= ⊥

⊥⊥⊥

⊥ νμ

μν

hgAAmmA

ghmH B

IIB

I22

||2

463)( (3a)

⎟⎟⎠

⎞⎜⎜⎝

⎛⎟⎠⎞

⎜⎝⎛ −−−⎟

⎟⎠

⎞⎜⎜⎝

⎛= ⊥ ν

μμν

hg

AmmAghmH B

IIB

I||22

||||

|| 463)( (3b)

where h is the Plank’s constant, ν the microwave frequency, and mI is the nuclear magnetic

quantum number (mI= ±7/2, ±5/2, ±3/2, ±1/2). From the fit of the data, we obtain A|| and A⊥,

presented in Table 2 along with bibliographic results to compare [22].

Our EPR results in non-doped VOx NT’s are completely different respect to those reported

by Kweon and co-workers [23], where they fitted their data assuming two wide resonance lines

to explain their experiment. One of them presents similar satellites as those observed by us.

However, their second broader (~ 750 G) resonance line associated to the absorption of V4+-V4+

dimmers is not present in our case.

The doped samples present more complex spectra because, in addition to the V4+ lines

complexity, they also present the resonance lines provided by the magnetic doped ion.

Our first studied case is the EPR in X-band of Ni-doped NT’s (Fig. 6), where two

contributions are easily visible. These contributions are described by the V4+ and Ni2+

Hamiltonians ( SHg NiVNiV .22 ++++ +=+= 44 HHHH ). In the spectrum both contributions

are presented; the first is a signal with a superstructure centered at g≈1.96 with a well resolved

hfs. This signal corresponds to the V4+ ions. The other broad line (~ 1400 G) is well described

by a Lorentzian lineshape centered at +2Nig ≈2.22 which is in agreement with the expected

value for Ni2+ paramagnetic ions [24]. In order to adjust the EPR parameters of the Ni

contribution we fitted the spectrum keeping the width and center field as free parameters. The

fitted EPR parameters for both contributions are presented in Table 2; the g’s and A’s factors of

the V4+ contribution were calculated using Eqs. 3.

The Co-doped vanadium oxide NT’s are a more complex case because we can not explain

the behavior with only two components for the two magnetic ions (V and Co). To make a very

good description, we need to introduce for both ions the g|| and g⊥ contributions (crystalline

anisotropy). The Co EPR resonance can be calculated considering quantum transition

probabilities between the energy levels, the crystal field effect on these levels and its angular

7

distribution. To describe the asymmetric Co resonance, we use a powder line-shape to fit the

spectrum for each employed microwave frequency.

To obtain the position of the resonance field for each angle, we have to solve the following

Hamiltonian [25]:

)}(sin21)(sincos

)]1(31)[sin{(cos

])cossin(cossin)[(

222

222

2||

2||

yxxzzx

z

zxBeff

SSSSSS

SSSD

SggSggH

−−++

+−−

+++−= ⊥⊥+

θθθ

θθ

θθθθμ2CoH

(4)

where Heff = H-H0 is the effective magnetic field with H0 being an internal field and H the

external magnetic field; D is the orthorhombic crystal field parameter and θ is the angle between

the z direction of the crystal and the external magnetic field (θ = 0º, parallel and θ = 90º,

perpendicular).

Finally, the Co line (Y(θ,H0)) was calculated by adding each angular contribution weighted

by the angle position (sin θ):

∑ ∑=

≠

=θ

θθθθ sin),,(),(),( 0

4,3,2,1,

0 HHfHIHY ijijij

jiji

(5)

where f(θ,Hij, H0) is the Lorentzian line describing the absorption as a function of the polar

angle (θ), the resonance field (Hij) and the internal field (H0); Iij(θ,Hij) is the intensity of each

transition and angle, calculated from the eigenvectors of the Hamiltonian matrix. Hij is the

magnetic field where the difference between two energies (Ei and Ej) is equal to hν/μB. The

energy levels are the eigenvalues of the Hamiltonian matrix

In Fig. 7a and 7c we show the magnetic field dependence of the energy levels for the two

extreme cases, θ = 0º and θ = 90º. In Fig. 7b, the magnetic field dependence for intermediate

polar angles is represented. The calculated intensities (Iij) of the low field transitions (1-3, 1-4

and 2-4, see left vertical lines in Fig. 7) are negligible, while the high field lines contribution to

the total resonance is much stronger. The envelope curves, for the three frequency bands, can be

well fit with the proposed cobalt model (see solid line in Fig. 8a, 8b and 8c).

The superstructure of the V4+ was described by both terms of +4HV

(Eq. 2). We fit the

position of the hfs resonances by Eqs. 3a and 3b. In the inset of Fig. 8a we show the

experimental and calculated magnetic field position (Hi) for parallel and perpendicular hfs

vanadium lines as a function of mI. On the other hand, in the inset of Fig. 8b, we plot the

residual experimental data after subtracting the model for the Co contribution (fitted by Eq. 5).

The upper and lower bars in this inset indicate resonance fields of the perpendicular and parallel

8

contribution of the hfs of the vanadium ions and we show how these bars coincide with the

position of the resonances in the spectrum. The fitting parameters are shown in Table 2.

We remark that Co-doped NT’s and Ni-doped NT’s systems, after the subtraction of the

doping ion resonance, show the same residual lines in amount and magnetic field positions

which can be directly associated to the hfs contribution of V4+ ions. Indeed, 58Ni (68.0%

abundance) and 60Ni (23.2% abundance) with I=0 do not present hfs resonances. On the other

hand, although 59Co has nuclear spin (I=7/2) and approximately a 100% of abundance, no hfs

corresponding to these atoms was detected du to the exchange narrowing effect. These features

constitute good evidence that the observed residual lines are only produced by the hfs 51V in

both doped samples (Co and Ni). Also, we can assert that the observation of the well-resolved

V4+ hfs resonances indicates: i) the presence of isolated V4+ magnetic ions and ii) that these

doped samples have a considerable smaller V4+ amount than the non-doped VOx NT’s.

3.3. dc-magnetic characterization

The dc-magnetization as a function of temperature of non-doped and doped NT’s measured

with an applied field of 10kOe is presented in Fig. 9.

In a previous work we have described the magnetic behavior of Ni-VOx NT’s as a

paramagnet which follows the law: χ(T)=C/(T-Θ)+ χ0 (Curie-Weiss model that a temperature

independent term), where C is the Curie constant, Θ is the Curie’s temperature, χ0 is a

temperature independent contribution, which has been calculated following the procedure

previously described in Ref. [26]. In the studied temperature range the samples did not present

antiferromagnetic dimmers, as was reported by other authors [5, 6]. In the Ni and Co doped

samples we assume that C has two contributions: one from the paramagnetic doping species

(Ni2+ or Co2+) and the other from to the V4+ ions. As the experimental results were normalized

by the moles of V, the units of C are expressed in moles of V too. Them the Curie constant can

be written as:

∑+++=

+=224 ,,

/

22

).1(3

..

NiCoViViii

B

iBA FSSk

gNC

μ, (6)

where kB is the Boltzmann constant, <gi> is the mean value of the g-factor, Si is the spin of the i-

th magnetic ion and; Fi/V is the ratio i-mole/V-mole. From the Curie constant we calculated the

V4+ percentage in the non-doped NT’s to be 16(2)%. Alternatively, for the doped case, the

amount of V4+ can not be calculated directly from C and to obtain this we need to subtract first

the contribution from Co and Ni. To calculate the Co and Ni contributions to C, we take into

account our EPR results: the ions are present as Co2+ (3d7; S=3/2) and Ni2+ (3d8; S=1) and their

9

<g>-factors correspond to these in Table 2. Those reported considerations allow us to estimate

the V4+ amount in the NT’s and the results are presented in Table 3.

To verify the amount of V4+ obtained for the three samples from the temperature experiments

we performed measurements varying the magnetic fields at low temperatures. The M vs. H/T

curves do not follow a linear trend (Fig. 10) because our diluted paramagnetic ions are partially

saturated by H/T. However, we obtained a very good fit with the model presented in Eq. 7. In all

cases we have to take into account two terms: the linear contribution (second term in Eq. 7) and

a Brillouin function [27] corresponding to V4+ ions. When we consider the Co2+ or Ni2+ case, we

need to add an extra Brillouin contribution for the transition metal doped ion. Them

magnetization can be described by:

( ) HFxBMTHM ViNiCoVi

iSi

i...)/( 0/

,,0224

χ+= ∑+++=

(7)

where M0i= gi SiμB, BS(xi) with xi=giμB.Si.H/(T.kB) is the Brillouin function and its argument that

describes the paramagnetic behavior of i=V4+ , Co2+ or Ni2+, as appropriate. The parameters

obtained from the fitting of the M vs. H curves are presented in Table 3.

The two independent magnetic experiments (varying the temperature and varying the

magnetic field) in the non-doped NT’s are consistent. In both cases we obtained between 14 and

16% of quasi-free spins, which are responsible of the Curie-Weiss behavior observed in the

magnetic susceptibility and magnetization. This result is also in good agreement with the

vanadium percentage that occupies the V(3) sites, 1/7 of the total vanadium sites present in the

unit cell. On the other hand, the V(3) are magnetically uncoupled because they are practically

isolated from their vanadium neighbors. Both our quantitative experimental EPR and

magnetization results of the quasi-free V4+ ions and those reported by Krusin-Elbaum [5] and

Vavilova [6] in similar VOx NT’s, suggest that this percentage is a general characteristic of the

VOx system and this not depend on the type of amine located between the oxide planes.

4. Conclusions

We observed in both the doped and non-doped samples that a fraction of the V ions that

constitute the VOx NT’s is in the V4+ state. This fraction is reduced significantly in the Co and

Ni doped samples which present sharp residual hyperfine V4+ EPR resonances that are due to

isolated V4+ ions. In these samples a broad line is also detected, corresponding to the doped

metal ion in each case. The broad Co-doped line is well described by a model which considers

the powder resonance of Co2+ ions with S=3/2 in an axial angular dependence of the cubic

crystal field, perpendicular and parallel g-factors and Zeeman contribution. On the other hand,

10

the broad Ni-doped line can be described by a single Lorentzian line of Ni2+ ions with

<g>=2.22.

The dc-magnetization studies (M vs. H curves), confirm the presence of S=3/2 coming from

Co2+ ions (in Co-doped NT’s) and S=1 corresponding to Ni2+ (in the case of Ni-doped samples).

In addition the dc-magnetization allowed us to quantify the percentage of paramagnetic V4+ ions

present in both samples. The VOx/Hexa NT’s have a fraction of V4+ between 14-16% while in

the Co-doped sample it is reduced to 5-2% and close to 4-2% for Ni-doped NT’s. This result is

in good agreement with the observable 51V hfs in both doped NT’s.

Finally, we have presented an alternative method for doping vanadium

oxide/hexadecylamine nanotubes with Co2+ and Ni2+, which allows to reduce the concentration

of V4+ ions keeping the tubular structure. The presence of these dopants can lead to an

interesting modification in the nanotubes in order to minimize the V4+ amount which is known

to reduce the performance of this material when it is used as a cathode in Li batteries.

Acknowledgements. MES acknowledges to CONICET for the studentship. HET and RDS are

members of CONICET. This work was partially funded by the following projects: in Argentina

by U.N. Cuyo 06/C203; ANPCyT (PICT-2004 21372, PAV and RN3M); in Brazil by FAPESP

(Proc. 03/10015-3).

References [1] C. N. R. Rao and A. Govindaraj, Adv. Mater. 21 (2009) 4208 – 4233. [2] F. Krumeich; H.-J. Muhr; M. Niederberger; F. Bieri; B. Schnyder and R. Nesper, J. Am. Chem. Soc. 121 (1999) 8324 - 8331. [3] M. Niederberger; H.-J. Muhr; F. Krumeich; F. Bieri; D. Günther and R. Nesper, Chem. Mater. 12 (2000) 1995 - 2000. [4] L. I. Vera-Robles and A. Campero, J. Phys. Chem. C 112 (2008) 19930 - 19933 [5] L. Krusin-Elbaum; D.M. Newns; H. Zeng; V. Derycke; J.Z. Sun and R. Sandstrom, Nature 43 (2004) 672 - 676. [6] E. Vavilova; I. Hellmann; V. Ketev; C. Täschner; B. Büchner and R. Klingeler, Phys. Rev. B 73 (2006) 144417. [7] X. Wang; L. Liu; R. Bontchev and A.J. Jacobson, Chem. Commun (1998), 1009-1010. [8] J.M. Reinoso; H.-J. Muhr; F. Krumeich; F. Bieri and R. Nesper, Helv. Chim. Acta 83 (2000) 1724 - 1733. [9] L.I. Vera-Robles; F.U. Naab; A. Campero; J.L. Duggan and F.D. McDaniel, Nucl. Intsr. Meth. Phys. Res. B 398 (2007) 534 - 537 [10] M.E. Saleta; J. Curiale; H.E. Troiani; S. Ribeiro Guevara; R.D. Sánchez; M. Malta and R.M. Torresi, Appl. Surf. Sci. 398 (2007) 333-336 [11] A.Azambre and M.J. Hudson, Materials Letters 57 (2003) 3005 - 3009. [12] M.E. Saleta; H.E. Troiani; S. Ribeiro Guevara; R.D. Sánchez; M. Malta and R.M. Torresi, J. Magn. Magn. Mater. 320 (2008) e268 - e271. [13] A.V. Grigorieva; A.B. Tarasov; E.A. Goodilin; S.M. Badalyan; M.N. Rumyantseva; A.M. Gaskov; A. Birknerc; Y and D. Tretyakova, Mendelev Comm. 18 (2008) 6 - 7.

11

[14] K.-F. Zhang; D.-J. Guo; X. Liu; J. Li; H-L. Li and Z.-X. Su, J. Power Sources 162 (2006) 1077 - 1081. [15] M.E. Spahr; P.Stoschitzki-Bitterli; R.Nesper; O.Haas and P.Novak, J. Electrochem. Soc. (1999), 146, 2780 - 2783. [16] A. Dobley; K. Ngala; S. Yang; P.Y. Zavalij and M.S. Whinttingham, Chem. Mater. 13 (2001), pp. 4382 - 4386. [17] R. Nesper and H.-J. Muhr, Chimia 52 (1998), 571–578 [18] B. Yasoda; R.P Sreekanth Chakradhar; J.L. Rao and N.O. Gopal, Mater. Chem Phys. 106 (2007), 33-38. [19] J.W. Wiench; C.J. Fontenot; J.F. Woodworth; G.L. Schrader; M. Pruski and S.C. Larsen, J. Phys Chem B 109 (2005) 1756-1762 [20] M.F. Deigen; V. Ya Zevin; V.M. Maevskii; I.V. Potykevich and B.D. Shanina, Sov Phys Solid State 9 (1967), 773-782. [21] R. Muncaster and S. Parke, J. Non-Cryst. Solids 24 (1977), 399-412 [22] A. Kahn; J. Livage and R. Collongues, Phys. Stat. Sol. (a) 26 (1974), 175-179. [23] H. Kweon; K.W. Lee; E.M. Lee; J.Park; I-M. Kim; C.E. Lee;G. Jung; A. Gedanken and Y. Koltypin, Phys. Rev. B 76 (2007) 045434. [24] K.D. Bowers and J. Owen, Rep. Prog. Phys. 18 (1955), 304-373. [25] C.P. Poole Jr. and H.A. Farach, The theory of magnetic resonance, (Wiley-Interscience, NY, 1972) [26] M.E. Saleta; J. Curiale; H.E. Troiani; S. Ribeiro Guevara; R.D. Sánchez; M. Malta and R.M. Torresi, Appl. Surf. Sci. 254 (2007) 371 - 374. [27] J.S. Smart, Effective field theories of magnetism, (W.B. Saunders Company, USA, 1966) Chapter 1.

12

Tables

Table 1: percentage of V and Co atoms measured by INAA and metal/V ratio calculated from

EDS data.

% V (INAA)

% metal

(INAA)

metal/V ratio

(EDS)

d distance (XRD)

Non-doped VOx/Hexa 37.4 (9) - - - - - - 2.1

Co-doped VOx/Hexa 30.9 (5) 4.2 (3) 0.14 1.2

Ni-doped VOx/Hexa (a) 32.6 (7) - - - 0.16 1.3

(a) previously reported in [11]

Table 2: Principal values of the hfs EPR spectrum measured at room temperature. The values for a V2O5 single crystal and amorphous were reported in Ref. 23.

Metal V4+ g|| g⊥ <g> G|| g⊥ <g> A|| [G] A⊥ [G]

Non-doped VOx/Hexa 1.93 (1) 1.96 (1) 1.95 188 (5) 92 (2)

Co-doped VOx/Hexa * 1.98 (2) 1.95(2) 1.96 1.930(5) 1.980(5) 1.963 185 (5) 74(1)

Ni-doped VOx/Hexa 2.22 1.930(5) 1.980(5) 1.963 192(5) 79(4)

V2O5 single crystal [23] 1.923 1.986 1.965 187 67

V2O5 amorphous [23] 1.926 1.984 1.965 211 79

* H0 = -483 (5)G and D= 433 (20) G.

Table 3: Fitting parameters obtained from the magnetic χ vs. T and M vs H/T curves.

Magnetic susceptibility curve (a) M vs H curve C

[emu.K/V mole Oe]

θ [Κ] %V4+ %V4+ Metal/V (R)

Non-doped VOx/Hexa 0.0606 (2) -3.90 (4) 16 (2) 14 (1) - - -

Co-doped VOx/Hexa 0.269 (9) 2.26 (6) 4.6 (4) 2 (1) 0.12 (1)

Ni-doped VOx/Hexa 0.2128 -1.74 (5) 4 (2) 2 (1) 0.13 (1)

(a) we assume the Co/V ratio measured by INAA and EDS (0.14 for Co and 0.16 for Ni). In a

previous work [11] we have already presented preliminary results of the magnetic

characterization for Ni-doped NT’s.

13

Figures Captions

Figure 1: Structure of the VOx layer, the V(1) and V(2) sites have octahedral coordination,

while V(3) sites are in the center of a tetrahedral of oxygen ions. (a) Detail of the V(1,2)-O

double layer. Between them are the V(3) sites, where vanadium atoms are less magnetically

coupled with other vanadium atoms. (b) The same schematic array of atoms showing the

polyhedra structure that forms the environment of oxygen around the vanadium atoms.

Figure 2: TEM micrographs of an uncompleted formed (a) Ni-doped NT and (b) non-doped NT.

(c) and (d) schematic representation of the rolling-up process of the NT’s.

Figure 3: (a) TEM image of a non-doped NT. (b) EDS spectra of non-doped and doped NT’s,

the Cu lines as produced by the sample holder. (c) Low angle (left) and high angle regions

(right) XRD patterns for the non-doped and doped samples

Figure 4: X-band EPR spectra of VOx/Hexa, Co-VOx/Hexa and Ni-VOx/Hexa NT’s recorded at

room temperature.

Figure 5: EPR spectra of VOx/Hexa NT’s recorded at room temperature, the experimental data

were fitted using a powder line shape for ions with S=1/2 (V4+). (a) X-band (9.5GHz), (b) K-

band (24GHz) and (c) Q-band (35GHz).

Figure 6: (Color online) EPR spectrum of Ni-doped NT’s collected at X-band and room

temperature. The Ni signal was fitted with a Lorentzian function to describe the absorption

curve; we also indicate in the upper and lower parts of the graph the positions of the V hfs lines

(small vertical lines).

Figure 7: (a) Energy levels along z (θ = 0º). Calculated transitions at Q-band are indicated by

vertical lines. (b) Angular dependence of the resonance magnetic field. (c) Energy levels

corresponding to the xy plane (θ = 90º) configuration, vertical lines indicate the resonances at

Q-band.

14

Figure 8: (color online) Room temperature EPR spectra of Co-doped NT’s collected at: (a) X-

band, (b) K-band and (c) Q-band. The noisy (black) curve is the experimental data and the

smooth solid (red) line is the calculated spectrum for the Co contribution assuming a powder

distribution. The inset in (a) shows hfs lines of the V contribution fitted with Eqs. 1a and 1b.

Inset in (b) is a detail of the hfs of the V ions.

Figure 9: (Color online) Magnetic susceptibility as a function of temperature (solid circles: non-

doped VOx/Hexa NT’s; open circles: Co-doped NT’s; stars: Ni-doped NT’s). Inset: (χ−χ0)-1 vs.

temperature. χ0 involves all diamagnetic contributions and its calculation is mentioned in the

text.

Figure 10: (color online) (Mmolar - χ0.H) as a function of H, for (a) non-doped VOx/Hexa NT’s,

(b) Co-doped NT’s, (c) Ni-doped NT’s. In all the samples the solid lines correspond to the fit

using Eq. 7 (see text). For both doped samples we also plot the contribution from each magnetic

ions.

15

(a)

(b)

Figure 1

16

Figure 2

17

Figure 3

18

0 2 4 6

EP

R s

igna

l [ar

b. u

nits

]

Magnetic Field [kG]

x 50 (Co-doped VOx NT's)

x 10 (Ni-doped VOx NT's)

x 1 (non-doped VOx NT's)

Figure 4

19

Figure 5

20

Figure 6

21

Figure 7

22

Figure 8

23

Figure 9

24

Figure 10